koadah
Joined: Mar 30, 2005
|
 Posted:
Apr 02, 2012 - 16:38 |
 
|
| SillySod wrote: | | VoodooMike wrote: | | The majority of games played on FUMBBL and on Cyanide (and thus, on online venues, which I'm fairly certain represent more games than tabletop over any given period of time) are done via TV matching in open play. That's the base premise I typically work from. |
I'm afraid to say that thats a bit of a doomed premise. TV is just an incorrect way to value a team so you are going to find yourself running into all kinds of difficulty.
| koadah wrote: | | Hands up all those who think CPOMB is undervalued. What about block, sure hands, accurate? Are we gonna whine about that too? |
We definately should. Moving people from claws onto amazons is not a win. |
Ain't nuffink wrong wiv Amazons. It's them stinkin lizardmen what's the problem. 
Taking tackle helps with them too.  |
_________________
[SL] + Official Stunty teams. Progression KO. Old & new teams welcome. 29th May! |
|
Mahvo
Joined: Feb 06, 2007
|
| Posted:
Apr 02, 2012 - 17:25 |
|
I have two ideas.
1) What about simple "unless you have a full roster of 16 players and at least 3 RRs (and apo if eligible), your treasury counts to TV"?
Usually grinders have loads and loads of cash. Hell Grannies is a prime example, 440k cash, no apo, 12 players.
People could dump their money, but that means that when that special blitzer dies, you might be replacing her with a Loner in the next match, then another star dies and soon your team is no more.
2) Calculate average (or median?) SPP of the three top stars in the team and make the pairings just based on that and let the inducements work out the rest. Forget the TV totally.
People with a lot of legends will face people with a lot of legends, people with rookies will face people with rookies. Problem solved.
Actually I quite like the second idea. It solves everything! |
|
|
VoodooMike
Joined: Nov 07, 2010
|
| Posted:
Apr 02, 2012 - 18:36 |
|
| JimmyFantastic wrote: | | I mean VM, have you done any statistical analysis of C-Pomb Min-Maxers? Are their winning % anywhere close to "fair" TV optimised teams such as Hell Grannies and Jungle Jingles? |
Nope, because the data is not presently available unless I want to bog down the servers with many days of data scraping, which I'm pretty sure neither myself nor Christer would be happy about. If we can all think up a better way to get more comprehensive data without it requiring many days of requests, I'm game to analyze anything!
That said...
| JimmyFantastic wrote: | It seems pointless to put in an anti min-max formula when they aren't the teams creating the most "unfair" games.
They are simply the teams that people whine about the most. |
Yep, and I'm focusing on what people have been whining about overwhelmingly during discussions of TV matching. I'm not going to try to make a game that is inherently unbalanced into a game that is perfectly balanced... but I can try to look for solutions to what make people less likely to enjoy themselves while playing the existing unbalanced game.
| JimmyFantastic wrote: | | 103 spps and no apo on a 1160 TV team. If that isn't a high development team crafted to win against low development teams I don't know what is. |
Roster problem, Jimmy. Look at how many games each player has actually played in! It's just a matter of Amazons being ridiculously good at low TVs against all but teams with "Dwarf" in their name. Really, if a team can become your definition of "minmaxed" simply by taking the first skill choice most people would suggest to a new coach for their first roll on any player... you're just saying that the team/race itself is unbalanced.
| Igvy wrote: | | min-maxing is defined (everywhere else but in this thread) as maximising one perticular ability at the expense of all else, with the hope that it covers for everything else. |
Maybe it's defined that way in other GAMES, but every time I've seen it referenced in BB related threads, it has been high development teams that are playing in a lower TV bracket against teams that can't deal with them due to the fact that the average amount of development at that TV is much lower... the "min" is minimizing TV, and the max is maximizing the effect of the team's roster.
| Hitonagashi wrote: | | So, I need to use the 1 pass and sack if mvp system? That still doesn't change what I have to do...these players are taking hundreds of hours to develop these teams, most of the clawbomb legends have taken 50-70 games each to get to where they are. |
Really, you need to run the numbers instead of just imagining that the things you say will work themselves out. Lets look at the premise you'r so proud of here:
1) Gaining SPPs through any means other than MVP comes from taking risks that could lead to turnover, and you'll be doing it with rookie players. The process of getting those SPPs involves turn-ending attempts with crap players.
2) Someone is going to get MVP, and you'll have to axe them and then go take more risky actions with a new rookie over and over until they get the required amount of SPP (but not too many!)
3) Even if you manage to do all this, you're still not going to end up avoiding a TV increase... lets look at the two teams used in the original example, and what would happen if their rookies all had, say, 5 SPP:
Zzzaifjtopioz TV 1380 -> TV 1685 -> TV 1569
New Colibrì Ruzzolanti TV 1380 -> TV 1885 -> 1490
So, does going through all that work decrease the handicap? Yep, but it doesn't eliminate it - the teams are still being pushed upward in matching TV nomatter how they decide to bust their asses to game the system. As I said.. it's not knife-edge perfect, but it serves its purpose just fine, even with your theoreticals (that you're not checking yourself).
| Hitonagashi wrote: | | If in each of those games, I made a single pass, that's easily most of my team on 1 spp, and I bet I could get your median down to 1 (2 legends, 3 people with one pass? I'd say that's pretty darn easy). |
The system uses a minimum Median of 1, so if the median SPP of the players is 0 then it is treated as 1 simply because you can't divide by zero. Give all your players 1 spp and nothing has changed. What you really want to do is give all your players the maximum number of SPPs you can have without changing the player's contribution to TV.
| Hitonagashi wrote: | | At the same time, those who are skilling their team evenly actually get targeted by this, as if we take a team with 5 players with skills, a few 2 skills and a few 3 skills, and the rest rookies (say, a 5 game old team), the median will be 2 or 3 TV. Compare this to our one pass clawbomb team of previously equal TV, and suddenly, they get paired against even lower teams! |
Again, while I'm sure this sounds logical in your head, you need to run the numbers. The number of SPP needed to get a skill roll increases with each level, so new teams, even when they skill up, will have low mean SPPs even with the setup you're talking about. It's unlikely that they'll manage to get the mean/median calculation to 2 or higher, and if they do, it will be negligibly higher than 2, meaning that the adjustment will be less than 1% of their TV.
| Hitonagashi wrote: | | You are building a system that only works given the assumption that people will not try and game it. The existence of the problem itself is proof this is wrong. |
No, you're attacking the system using the assumption that your feelings will translate into numbers effectively. Numbers where your mouth is, my friend!
| Nelphine wrote: | You seem to want to stop 'legend+lino' teams, at low TV, because they are too successful - but you don't care why they get there.
But why should that matter IF they are no more successful than the amazon team? (My first example) |
I want to give a way to adjust their TV upward to place them more appropriately (or at an inappropriate disadvantage, really). I don't care how they got there, it's true, and I know that they're not all successful at the same thing. Some are winning monsters, some just rip people up.
I don't know why it matters, but it seems to. Other than CLPoMB ranting, it is the most frequent complaint I see related to TV matching.
| Nelphine wrote: | | On the other side of things: Imagine a pro elf team with a legend thrower and (due to a few bad games) a team full of rookies. Your system will penalize that team MORE than it will penalize an orc team with several block skills a few guard skills and perhaps a dirty player. (my second example) Yet (in my opinion) that thrower will NOT provide any particularly greater advantage than the 6 skills on the orc team. So your system mistakenly gives the elf team a 'handicap', even though the orc team (without your system) is probably better off. |
the orc thing is a roster issue, if its an issue at all. The first one will treat the pro elf team as being a higher TV than it actually has, for matching, yes... and at that point, for that team that has lost everyone but a thrower, the thrower is pure TV bloat because he's contributing more to TV than he's contributing to the team's success. If you hold on to him in the hopes that he'll be useful SOMEDAY even though you know it won't be any time soon, you've made peace with inflated TV anyway.
| Nelphine wrote: | | This is why I DO believe amazons are a minmax team. (Dorfs are too, but to a lesser extent; at low TV, dodge is superior to block, due to the lack of tackle and block on most low TV teams) Because they win more games. |
Right, and this is why I say you're lumping your roster issues into the conversation the same way people like to try to turn every conversation into a CLPoMB rant. The solution to the problem you see in Amazons is to adjust the cost of amazon players, or to change the skills they have, both of which will involve ROSTER changes. It is a ROSTER issue.
If you take no positionals then that's your choice, because hey, teams can have 0 of a given player type on them and often do have 0 of a given positional... people only take them because they USUALLY have more bang for the buck than linemen.
| dode74 wrote: | Regarding adjustment, looking at the team's win% and adjusting according to equivalent TV-difference win% as calculated from the data we have on TV differences and win%s. Mike previously came up with the following as a simple linear regression:
(UD win ratio) = 0.458 - 0.003 x (rating difference)
Adjusting that, we could make the "rating difference" the TV matching adjustment, such that:
(TV matching adjustment) = (Win% - 0.458)/0.003 |
Remember that I started that post by saying it was based on aggregation which was a bad idea, and that the line accounted for insufficient variance to be considered useful, but here was the regression equation anyway...
| dode74 wrote: | | Clearly, this is matching by record rather than directly by TV, but that does appear to be what is wanted for minmaxers (among those who want to do something about it). |
The main problem is that it will also apply handicaps to races that are good at certain TVs, assuming we can maximize the variance accounted for by the equation... so a fresh 1000 TV amazon team will still get matched up with, say, 1200 TV khemri, or whatever. I'm not sure that's a good solution to anything, but hey, maybe I'm in the minority?
| SillySod wrote: | | I'm afraid to say that thats a bit of a doomed premise. TV is just an incorrect way to value a team so you are going to find yourself running into all kinds of difficulty. |
It is, however, the way that most matching is happening in most Blood Bowl games on our planet, so... y'know... I'm going to work with what we have, because nobody has thought up a better system yet. Lots of handwaving, yes, but nothing concrete.
| Koadah wrote: | Eh? So do you want to match by min/max quotient?
Was it still April 1st when you wrote that? Shocked |
The "you must be joking" thing is getting old now, koadah. I want to match by TV (mostly because I havent seen a better system yet) and adjust the TV upward if the teams meet a very easy to calculate criteria. Most of my solutions to pretty much everything involve coming up with PLAUSIBLE and COMPLETE ideas that are easy to implement, rather than nebulous bullshit that someone else has to try to solidify through hours of their own work.
| koadah wrote: | | We're not talking about maths. We're talking about a game. The maths doesn't really matter. What matters is "do people like it". The numbers you need to be looking at is how many people like it. How many don't. How many can live with it. How many can't. |
It's a game of numbers and math, so math matters... and as it happens, I AM looking at things in terms of what people feel.. I'm just not presenting the idea of a panacea, and people's objections seem to be that its not a panacea... seems to be people's objections to everything anyone says. "Well it doesn't solve world hunger, so there's no point"...
| mahvo wrote: | | 2) Calculate average (or median?) SPP of the three top stars in the team and make the pairings just based on that and let the inducements work out the rest. Forget the TV totally. |
You'd run into problems at high TVs when bashy teams start meting out permanent casualties, but you can't stop getting paired with them because you haven't lost your final 3 decent players. It makes it similar to pairing by games played. |
|
|
Purplegoo
Joined: Mar 23, 2006
|
| Posted:
Apr 02, 2012 - 19:07 |
|
Blimey, a long weekend away and this goes bananas?
Am I right in thinking I don't need to read it and it's the same old stuff with a Voodoodriver? |
|
|
licker
Joined: Jul 10, 2009
|
| Posted:
Apr 02, 2012 - 21:22 |
|
Same old argument, new thoughts on a solution to it.
Read it or don't, nothings changing anyway is it? |
|
|
Mahvo
Joined: Feb 06, 2007
|
| Posted:
Apr 02, 2012 - 22:08 |
|
| VoodooMike wrote: | | mahvo wrote: | | 2) Calculate average (or median?) SPP of the three top stars in the team and make the pairings just based on that and let the inducements work out the rest. Forget the TV totally. |
You'd run into problems at high TVs when bashy teams start meting out permanent casualties, but you can't stop getting paired with them because you haven't lost your final 3 decent players. It makes it similar to pairing by games played. |
Well. That's tough luck. Maybe those three decent players win the game for me. At least I'll have babes, wizard and what not to back me out, if situation is so bad.
Important thing is that you won't be facing noobie teams with three legends or vice versa.
I think that it is STILL a perfect solution. Maybe adjust the amount of stars used to calculate the median/average (top 5?) in order to make it more all around solution on high TV levels, but the basics are solid. |
|
|
Hitonagashi
Joined: Apr 09, 2006
|
| Posted:
Apr 02, 2012 - 23:25 |
|
@Purplegoo: Yep
| VoodooMike wrote: |
| Hitonagashi wrote: | | So, I need to use the 1 pass and sack if mvp system? That still doesn't change what I have to do...these players are taking hundreds of hours to develop these teams, most of the clawbomb legends have taken 50-70 games each to get to where they are. |
Really, you need to run the numbers instead of just imagining that the things you say will work themselves out. Lets look at the premise you'r so proud of here:
1) Gaining SPPs through any means other than MVP comes from taking risks that could lead to turnover, and you'll be doing it with rookie players. The process of getting those SPPs involves turn-ending attempts with crap players.
2) Someone is going to get MVP, and you'll have to axe them and then go take more risky actions with a new rookie over and over until they get the required amount of SPP (but not too many!)
3) Even if you manage to do all this, you're still not going to end up avoiding a TV increase... lets look at the two teams used in the original example, and what would happen if their rookies all had, say, 5 SPP:
Zzzaifjtopioz TV 1380 -> TV 1685 -> TV 1569
New Colibrì Ruzzolanti TV 1380 -> TV 1885 -> 1490
So, does going through all that work decrease the handicap? Yep, but it doesn't eliminate it - the teams are still being pushed upward in matching TV nomatter how they decide to bust their asses to game the system. As I said.. it's not knife-edge perfect, but it serves its purpose just fine, even with your theoreticals (that you're not checking yourself).. |
You've just made the point I was trying to for me. I didn't run the numbers, because I couldn't be bothered to spend the time to calculate them, and I could see them in my head. Thanks for doing it though...but your situations are exactly what I think is broken
In the case of New Colibri, that's approximately 395 TV drop for *no in game effect*. In normal TV measure, that works out as 19 skills...gee, if I could make 1 pass a game (and sack a rookie that gets MVP) to have a 20 skill edge on my opponent, I think I might just do that...
I don't know if you played LRB 4, but you seem to be under the impression that 1 pass a game is hard. It's really not. |
|
|
Vesikannu
Joined: Mar 06, 2011
|
| Posted:
Apr 03, 2012 - 07:05 |
|
| Hitonagashi wrote: | You've just made the point I was trying to for me. I didn't run the numbers, because I couldn't be bothered to spend the time to calculate them, and I could see them in my head. Thanks for doing it though...but your situations are exactly what I think is broken
In the case of New Colibri, that's approximately 395 TV drop for *no in game effect*. In normal TV measure, that works out as 19 skills...gee, if I could make 1 pass a game (and sack a rookie that gets MVP) to have a 20 skill edge on my opponent, I think I might just do that...
I don't know if you played LRB 4, but you seem to be under the impression that 1 pass a game is hard. It's really not. |
You'd have to do five passes on each of your players, right? Those calculations are for each of the rookies having 5 SPP. You did read what VM wrote next, right?
| VoodooMike wrote: | | Give all your players 1 spp and nothing has changed. What you really want to do is give all your players the maximum number of SPPs you can have without changing the player's contribution to TV. |
Is it still worth it? I know some people will do anything, but would games really turn into rookie-passing madness if this was implemented. |
|
|
Nelphine
Joined: Apr 01, 2011
|
| Posted:
Apr 03, 2012 - 07:18 |
|
I already do rookie passing madness; I've been doing upwards of 5 passes a game with my chaos team (albeit the number increased when I got my agi increase). |
|
|
Leilond
Joined: Jan 02, 2012
|
| Posted:
Apr 03, 2012 - 09:58 |
|
I never fire a linerat unless he recive a niggling or a permanent injury
The problem is that every single linerat ALWAYS recive one of those injury before they reach 16 spp and usually before 6
A normal skaven team is ALWAYS formed by 4 76+ spp gutter runners and 5+ rookie linerats... not for because we minmax, but because after 50+ matches, only one of my linerat survived (kick and +ST, for a grantotal of 19 spp)... never fired a single not injuried linerat.
Thus, a formula based on the "legend + lots of rookie" don't work well |
|
|
Ehlers
Joined: Jun 26, 2006
|
| Posted:
Apr 03, 2012 - 11:12 |
|
I have followed and not sure exactly what all this is about.
But so far, we dont want legend plus rookie teams right?
To solve this, we do
| Quote: |
Mean player SPP / Median player SPP (min 1)
If the result is 2 or higher, use the result as a percentage to increase the team's TV by. |
But is this system not broken toward teams that perform well with a few key players and lots of fodders/rookies? (skaven and nurgle just come to my mind)
The Rotters die faster on the pitch than I can spell pie.
But is this system not broken to teams who get miss next game players?
Most teams aim towards getting really good players, but you will in a lot of times get some players that a better than the others due to they are your ball carrier or just cas better than the rest and try in their best ability to skill up the rest of the team.
So what happen to a team which has a few better players than the rest (all players above 2 or 3skills) but suffer a really hard match. They will be left with a few legends and journeyman. Now suddenly they fall into your min/max formula and get matched against higher TV regardless if they did not min/max on purpose.
So my team get a heavy beating in black box, I hope for an easy match to get my team back on track, but instead will be paired with a higher TV and therefore more likely this team has more skills to make my team suffer more.
What is your take on this? |
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
| Posted:
Apr 03, 2012 - 11:23 |
|
And there is your problem VM, people whine about it because of the intention of it, not the effectiveness of it.
Best just to ignore the whiners. |
_________________
Pull down the veil - actively bad for the hobby! |
|
koadah
Joined: Mar 30, 2005
|
| Posted:
Apr 03, 2012 - 12:08 |
|
|
harvestmouse
Joined: May 13, 2007
|
| Posted:
Apr 03, 2012 - 13:07 |
|
| JimmyFantastic wrote: | And there is your problem VM, people whine about it because of the intention of it, not the effectiveness of it.
Best just to ignore the whiners. |
Someone's been whining a lot in this thread, don't think he could be classed as an anti minmaxer though! |
|
|
dode74
Joined: Aug 14, 2009
|
| Posted:
Apr 03, 2012 - 13:51 |
|
Isn't the "problem" simply that certain builds do much better than people think they should at certain TVs? |
|
|
|
|
| |
1150 - OWA TT Tourna...
Search
